Identification of manganese efficiency candidate genes in winter barley (Hordeum vulgare) using genome wide association mapping

Experimental design

Chl a fluorescence phenotyping

Chl a fluorescence was used to screen all experiments for Mn efficiency. Field and greenhouse fluorescence measurements were conducted using a handheld Plant Efficiency Analyser device (Handy PEA, Hansatech Instruments Limited, King’s Lynn, UK). Due to seasonal variations and operational limitations, barley plants were measured at different plant development stages, from the three leaves stage to end of tillering (Table 1). Chl a fluorescence is affected by chilling and freezing temperatures [44], thus screening was performed at temperature above 5 °C. For each line in the field or pot in the greenhouse, three plants were measured using the youngest fully emerged leaves. After 25 min of dark-adaptation using Hansatech leaf clips, fluorescence measurements were recorded for 2 s after illumination with a saturating light pulse of 3000 μmol photons m⁻²s⁻¹ on the adaxial leaf surface. Fluorescence transients (Fig. 1) and JIP-test parameters [45] were extracted using PEA Plus Software (v1.10). The JIP-test parameters used for the study are summarized (Table 2). The maximum quantum efficiency of PSII (F_V/F_M) was used as the main stress indicator [46] as it has been demonstrated to be sensitive for Mn deficiency diagnosis [3, 6, 21]. However, Mn deficiency leads also to other marked changes in the transients [21]. Therefore, besides F_V/F_M, also the parameters V_I, V_J, V_K and Area were analysed as the main parameters for association genetic analysis (Table 2). A full Mn-containing design as control was not realized. Indeed in regards to the PSII and the quantum yield efficiency parameter measured (F _v /F _m ), no diversity could be observed. Under those conditions, the quantum yield efficiency of PSII will reach its maximum with a value of F _v /F _m around 0.82. Instead of such Mn-containing design, control pots (or lines) were sprayed with a solution of manganese sulphate (MnSO₄) at 0.1 % of Mn containing two droplets of Tween® 20 detergent per litre as described by Schmidt et al. [6], in the field and greenhouse experiments. Plants were measured two days after spraying.

Extracted and technical fluorescence parameters
F0	=	F50μs, fluorescence intensity at 50 μs
FK	=	F400μs fluorescence intensity at the K step (at 400 μs)
FJ	=	F2ms, fluorescence intensity at the J step (at 2 ms)
FI	=	F30ms, fluorescence intensity at the I step (at 30 ms)
FM	=	Maximum fluorescence intensity
FV	=	(FM – F0), variable fluorescence from a dark adapted leaf
FV/FM	=	Maximum quantum efficiency of PSII photochemistry.
tFM	=	Time to reach FM, in ms
VK	=	(FK – F0) / (FM – F0), relative variable fluorescence at 400 μs
VJ	=	(FJ – F0) / (FM – F0), relative variable fluorescence at 2 ms
VI	=	(FI – F0) / (FM – F0), relative variable fluorescence at 30 ms
Area	=	Area under the curve between F0 and FM

Genetic analyses

Bioinformatics on candidate genes

Selection of environments

Six different environments were used in the study to screen the 248 varieties of the winter barley collection. Mn deficiency induction cannot be easily controlled in fields and soil pots. Environmental effects such as temperature, soil humidity and redox states, all have major impacts on Mn plant availability and consequently on Mn deficiency development [66]. Consequently genotype-by-environment interactions were expected.

Therefore to study these interactions, genetic determination of the F_V/F_M related trait, also called heritability (H²), and genetic correlation were calculated (Table 4). H² characterizes the variance of the trait captured by the genotypic effects relative to the total phenotypic variance. The two field sites ST12 and LJ12 showed very low H², at 0.06 for both whereas H² was up to 0.31 for the greenhouse experiment GH13A. The low H² estimated show that only a minor fraction of the total variance is due to genetic effects. In addition, when looking at F_V/F_M genotypic estimates in the 6 environments (Fig. 2), the response range to low Mn availability ranged from 0.38 to 0.71 in LJ13 whereas LJ12 had F_V/F_M values from 0.72 to 0.76 and ST12 from 0.66 to 0.74. Indeed LJ12 and ST12 were phenotype in spring whereas the other environments were screened in winter. The winter 2012/13 was too wet and mild to induce Mn deficiency, while it is induced mostly under dry and cold conditions. Consequently, because of low H², narrow variability and inadequate phenotyping time, it was decided to discard ST12 and LJ12 from the GWA analysis.

Environments	GH13A	GH13B	KS11	LJ12	LJ13	ST12	H2
GH13A	1						0.31
GH13B	0.61	1					0.26
KS11	−0.02	−0.11	1				0.18
LJ12	0.14	0.16	0.71	1			0.06
LJ13	0.27	0.28	0.16	0.25	1		0.23
ST12	0.11	0.14	0.24	0.84	0.2	1	0.06

Calculation of genetic correlations across environments (Table 4) provided a better understanding of how genetic control of Mn efficiency is related across the different environments. The weak genetic correlations confirmed that the Mn efficiency trait had differential genetic determination depending on environments also named genotype × environment interactions. These contrasts across environments can be explained by several factors including climate (solar irradiation, soil humidity and temperature) and physiological age of the plants relative to the time of Chl a measurements. Indeed, the severity of Mn deficiency is subject to major fluctuations over time and consequently it was also expected that the genetic variability among the varieties would be affected. Moreover, the ranking of variety’s F _v /F _m performances fluctuated over the environments and no grouping of variety in regards to their origins or row-type was observed.

Assessment of plant response to Mn deficiency

To validate the induction of Mn deficiency, plants in soil pots and in fields were sprayed with soluble Mn and measured two days after spraying. In all four environments, a clear impact of MnSO₄ foliar application was recorded on Chl a fluorescence curves and F_V/F_M values (Fig. 3). The fluorescence yield F_V/F_M of plants treated rose to a higher level (between 0.75 and 0.83) whereas the control plants remained at lower levels (between 0.47 and 0.66) (Fig. 3 b), confirming that the plants were indeed Mn deficient, but without any visual leaf symptoms (latent Mn deficiency). Therefore the ability to fully restore F_V/F_M to its maximum after foliar application of Mn sulfate confirmed that Mn is the only abiotic stress parameters influencing the F_V/F_M values under the conditions prevailing at the field sites and in the green house experiment. The J and I steps of the transients were flattened and less defined, and in severe cases a notable K-step was developed (Fig. 3 a). The most severe Mn deficient conditions were seen in the GH13B experiment in which Mn deficient plants showed flat J and I slopes. Furthermore, when Mn deficient plants in LJ13 were sprayed with MnSO₄, the Mn concentration in leaf tissues increased significantly (Fig. 3 c). Therefore, spraying of Mn salts confirmed that Mn deficiency was induced and that plants can fully recover the maximum range of F_V/F_M. Moreover, the ICP-OES quantification of other nutrients indicated that their concentration in leaf tissues remained stable (Additional file 2). Consequently only Mn was responsible for the change of F_V/F_M.

Based on the differences in F_V/F_M, on the shape of the curves and on Mn leaf tissue concentration, it was evident that plants were exposed to Mn deficiency.

Grouping from the population structure

The structuration analysis allowed us to distinguish four groups (Fig. 4). Two groups of two-rows from mid-northern Europe were identified. One group of six-rows and one group with a mixture of six-rows and two-rows from a southern European origin were also detected. In blue, the group with two-rows contained most of the breeding lines of Scandinavian origin. It confirmed the significance of adding population structure into the GWA model.

GWA results

Taken together for the four environments considered, 54 SNP markers were found to be significantly associated with the F_V/F_M trait (P-value < 0.001) and were distributed over 6 chromosomes. Of these, 11 were unmapped SNPs with missing annotations. Marker-trait association displayed distinct peaks corresponding to QTLs (Fig. 5). Other associations were detected for all JIP parameters and are discussed below. Despite strong genotype × environment interactions, a meta-analysis including all environments (data not shown) resulted in the identification of the same variants. However it does not reflect the reality as most of the associations are environment-specific. Different SNPs were detected in each environment (Fig. 5) implying that marker-trait associations are severely affected by the environment. In contrast, only four SNPs were identified using the Mn leaf concentrations (Additional file 3). Three markers constitute one peak on chromosome 6H and one marker on chromosome 5H.

Putative candidate genes

Quantification of Mn deficiency

Chl a fluorescence has previously been shown to be a powerful tool to quantify the severenes of Mn deficiency in plants [3, 6, 21]. However, little is known about differences in Mn uptake and Mn tissue concentrations in the plant tissue of varieties with contrasting Mn efficiency. Therefore, the quantification of Mn in leaves was used as an additional parameter.in the current study. Plants were clearly Mn deficient as indicated by the leaf tissue concentrations ranging from7-12 μg.⁻¹ Mn g DW [21].

However, the correlation, 0.59, between Mn concentration and F_V/F_M does not suggest a strong link between the two parameters. GWA did not reveal any strong association with tissue Mn as indicated by the absence of a correlation between leaf Mn concentrations and the functional pool of Mn in PSII using Fv/Fm values as a proxy.

Environment-specific QTLs

Mn efficiency appears to be a complex quantitative trait controlled by many genes, each contributing with minor effects. Genetic determination of the trait ranged between 0.18 and 0.31 in the four environments analysed (Table 4), which means that only 18 % to 31 % of the phenotypic variance was due to the genetic variance. This supports the variability of plant responses to Mn deficiency partly being determined by genetic background. However, in contrast to traits governed by a few loci with large effects, the variation in quantitative traits is caused by the segregation of multiple QTL with individually small effects that are sensitive to the environment [68]. Indeed, significantly associated SNPs were found with relatively small effects (0.011 to 0.039) for F_V/F_M (Table 6). Those markers are identified only within specific environments and the corresponding QTLs are consequently specifically related to a single environment. Collins et al. [13] introduced the concept of “constitutive” and “adaptive” QTLs controlling the response of crops to abiotic stress. A “constitutive” QTL is consistently detected across different environments whereas “adaptive” QTL is environment-specific. Stress sensitivity can therefore be due to the responsiveness of QTL regulation in specific environmental conditions or due to indirect causes contributing to the stress response. Staple or constitutive QTLs are naturally preferred in breeding.

In association mapping, detecting constitutive QTLs for abiotic stresses is challenging, since QTL effects are most likely small and controlled by one or several genes, and by epistasis and environments. Environment-specific QTL has been previously reported in cereals, in winter barley [69], in wheat [70, 71], and in rice [72]. In the case of the Mn efficiency trait, the four environments included, had varying characteristics due to various climates in the three different years, differences in soil conditions and/or different growth stages at the screening date (Table 1).

Moreover, poor genetic correlations between environments also support the environment-specific feature of the trait, with correlations in a range from -0.11 to 0.61 for the four environments analysed (Table 4). Nevertheless, with a genetic correlation of 0.61, the two greenhouse experiments imply that Mn efficiency in GH13A and GH13B was regulated by a set of shared genes under the controlled environmental conditions, with identical soil and comparable climates.

The results confirmed that environments have to be considered independently for the complex Mn efficiency trait. Consequently, different environmental conditions are needed to obtain a comprehensive picture of the trait and to identify the “adaptive” QTLs contributing to Mn efficiency.

Identification of candidate genes

Combining GWA, database screening and prior knowledge about the role of Mn in plants allowed several candidate genes to be identified, coding for PSI and PSII proteins, chlorophyll a/b binding proteins, GLP, Mn-SOD, and PP2C. Two different classes of encoding genes could be distinguished, the first related to photosynthesis and the others involved in various Mn dependent pathways.

In photosynthesis, reactions occur in the thylakoid membranes carried out by multi-protein complexes, namely ATP synthase, PSI, PSII, cytochrome b6f and light harvesting complexes. Proteins from these complexes were identified: chlorophyll a/b binding proteins, PsaA/PsaB, PsaE, PsaH and PsaN from PSI; Psb28, PsbN, PsbP, PsbQ and PsbW from PSII; and finally Lhc proteins in multiple copies at several genome positions.

Role of Mn in metabolic pathways

The light-harvesting chlorophyll a/b binding proteins belong to the Lhc family consisting of more than 20 different proteins and are connected with PSI and PSII [73]. Their primary function is related to the absorption of light through chlorophyll excitation and the transfer of the absorbed energy to photochemical reaction centres. Although Lhc are coded by well characterized and abundant CAB (chlorophyll a/b binding) genes [74, 75], the link to Mn metabolism is not trivial. Nevertheless, previous studies showed a reduction in Chl a concentration in Mn deficient plants and changes in the phosphorylated forms of LhcII have been observed [21].

Several protein subunits from PSI were also detected. Similarly to the chlorophyll a/b binding proteins there is no obvious and immediate links to Mn utilization. The subunits PsaA/PsaB are involved in electron transport and form the core centre of PSI [76]. PsaE is one of the three stromal subunits of PSI whereas PsaH and PsaN are specific plants subunits [77, 78]. Although Mn is not a PSI cofactor, a decrease in the PSI activity in Mn-limited conditions has been reported previously. The composition of the PSI complex is altered and a decrease in PsaA subunit is pronounced [79].

PSII consists of at least 20 protein subunits and catalyses a light-driven process, splitting water into protons and molecular oxygen. [75, 80]. The Psb28 protein is associated with the cytoplasmic side of the thylakoid membrane and known to interact with the core protein CP43 in the PSII monomer [81]. Even though no direct link with Mn has been demonstrated, recent studies have partly established its function. A Psb28 deletion mutant generated in Synechocystis showed slower growth rates and lower chlorophyll levels. [81, 82]. Recent studies suggest a role of Psb28 in the assembly and repair of PSII under heat stress [83, 84].

In barley, PsbP and PsbQ subunits are extrinsic proteins involved in the stabilization of the tetranuclear Mn cluster of OEC. [85, 86]. The crucial role of PsbP in Mn binding to PSII has already been established [87]. It has been described as a Mn storage protein delivering Mn²⁺ during PSII assembly. Its absence slows down the process of photoactivation. Whereas PsbP was found mandatory for OEC activity and showed a strong Mn dependence, PsbQ did not present a manganese-dependent activity [88]. However, recent studies established that PsbQ stabilizes PsbP binding and therefore contributes to the activity of the OEC in PSII complexes [89]. Finally, a study on the tobacco PsbQ mutant revealed the effect of PsbQ loss on Chl a fluorescence under low light conditions. For the PsbQ mutant, F_V/F_M dropped from 0.73 to 0.26 under low light conditions [90]. Hence, measuring Mn efficiency based on Chl a fluorescence appears to be linked to PsbQ activity. PsbN, a low molecular subunit encoded in the chloroplast, has exclusively been found in etioplasts, but its PSII function remains elusive [91]. Nevertheless, a recent study has demonstrated by reverse genetics in tobacco, that PsbN is involved in the biogenesis of PSI and PSII [92]. In addition, it has been demonstrated that PsbN is required for the repair and assembly of PSII [93].

Recent work with PsbW knock-out and antisense Arabidopsis mutants indicated an important role played by PsbW for the connection and stability of PSII-LHCII complexes [94]. Lower F_V/F_M values have been reported for mutants lacking PsbW, but despite the role of PsbW for the PSII complex, direct effects of Mn on PsbW remain unclear.

The identification of several candidate genes coding for PSII subunit proteins (PsbN, PsbP, PsbQ and PsbW) demonstrate that Mn is involved in the stability, in the assembly and repair of the PSII and in the loading of Mn into the photosystem machinery. Therefore, it suggests that some varieties are able to load Mn, assembly PSII and repair PSII better than other under Mn deficient condition.

Apart from photosystem related proteins, two other types of proteins were found to be involved in Mn efficiency: PP2C and GLP. PP2C is a well-known serine/threonine specific protein phosphatase. With 76 candidate genes identified in Arabidopsis [95], the PP2C family is considered the largest protein phosphatase family in plants. They are identified on several barley chromosomes (1H, 2H, 3H, 4H and 7H). These phosphatases are involved in the regulation of several signalling pathways and stress responses in plants [96]. In the current context, an interesting feature of PP2C is the relatively high concentration of Mn or Mg required to maintain activity. In moss Physcomitrella patens, the activation of PP2C occurs at 0.1 mM of Mn whereas more than 5 mM of Mg was needed to activate the protein [97]. It therefore suggests that Mn has a special feature in activating PP2C, which appears to be a factor influencing Mn efficiency.

The remaining genes identified by GWA are coding for GLP and Mn-SOD. GLP are often described in crops as having two main enzymatic functions, OxO and SOD, in response to biotic and abiotic stress [98, 99]. The two enzymes, Mn-SOD and OxO, generate hydrogen peroxide (H₂O₂), a signalling molecule in plant defence. OxO enzymes are localized in the apoplast and Mn-SOD in mitochondria; they work in antioxidant defence by enzymatic mechanisms scavenging reactive oxygen species (ROS). A first study demonstrated the involvement of Mn as a cofactor in barley OxO activity [26]. Hence, it has been confirmed by spectroscopy and crystallography that Mn is a required cofactor for OxO catalysis [100, 101]; moreover it revealed the Mn-SOD activity of GLP. GLP coding genes in barley are mapped on chromosomes 2H, 3H, 4H, 5H, 7H and classifies to five main subfamilies annotated HvGER-I to HvGER-V [102]. Finally, gene family studies on barley [103, 104] confirmed the OxO and SOD activity of GLP, their function in defence resistance, the duplication of GLP genes, their developmental stage and their tissue-specific expression. The excellent characterization of the barley GLP genes and activity may contribute to a better understanding of the role of Mn.